APPLICATION OF MULTIVARIATE ANALYSIS TO DEVELOP BOREHOLE WATER QUALITY INDEX MODEL IN IKONO AKWA IBOM NIGERIA

Abstract

Water quality are usually dependent/determined by many water parameters/variables thereby making water quality index determination complex and expensive if many parameters are to be taken into consideration. Therefore, the need to develop simpler WQI with few parameters which suitably represent all in considerations is imperative. 20 physicochemical parameters samples were collected, measures/analyzed and data were collected. Multivariate statistical analysis which includes hierarchical clustering analysis (HCA), principal component analysis (PCA), and multiple linear regression (MLR) are applied thus reducing the initial input 20 parameters to five proxy parameters and model coefficients were established. Resulting from these parameters dimentionality reduction by HCA AND PCA analyses, salinity, pH/turbidity, Cu, Hg and conductivity are the final five proxy parameters/ variables retained. Two borehole models WQIa and WQIb were developed with pH and turbidity respectively alongside the remaining four parameters.Borehole WQIa model has more autocorrelation than WQIb model as it has Durbin-Watson (DW value of 2.107 > 1.470). Comparative analysis of predicting power of borehole WQIa model showed that the model is the best as the adjusted R square ie coefficient of determination is very high (90.6%) compare to WQIb model with poor adjusted R of 14.5%. This implies that 90.6% of the variation in WQIa is explained by the predictors ie salinity, pH, Cu, Hg and conductivity while 14.5% of variation in WQIb is explained by the predictors ie salinity, turbidity, Cu, Hg and conductivity. This also showed that pH is a stronger predictor compared to turbidity as these two predictors differenciate the two models WQIa and WQIb independent variables. Similarly, WQIa model has high F value of 26.085 and DW of 2.107 which indicated best/normal. This model showed statistical significant correlation among the variables with P value of 0.000 compared with WQIb model with low F value of 1.440 and DW of 1.470 with no significant different as the P value read 0.308. Finally, the use of WQIa would make water quality routine monitoring and control rapid,less rigorous, low-cost, less biased and more objective to enhance sustainable development programmes in the study area.